Ricursive Intelligence的创始人Anna Goldie和Azalia Mirhoseini在TechCrunch Disrupt大会上展示了一套利用人工智能自动化芯片设计的方案1。该公司开发的AI系统能够自动执行芯片设计流程、从迭代中学习并持续改进设计方案1。
他们的AlphaChip系统发现了传统设计方法难以想到的非常规芯片布局，包括环形"甜甜圈"型和波浪形等创新设计1。最为突出的是，这套AI方案能将芯片设计周期从传统的六个月至两年大幅缩短至一周1。Goldie和Mirhoseini认为，面对未来五年计算需求可能增长至少10,000倍的挑战1，应该让AI承担执行工作，从而使人类团队能够专注于创意工作和应用开发1。
Ricursive Intelligence founders Anna Goldie and Azalia Mirhoseini unveiled their artificial intelligence-driven chip design approach at TechCrunch Disrupt, positioning the technology as a solution to the emerging compute crisis.1 The company's AI system automates the chip design process, learning from iterative cycles to generate unconventional layouts that traditional engineering methods would not produce, including ring-shaped and wave-patterned configurations.1 Their AlphaChip system has created distinctive "donut"-shaped chip designs, representing a departure from conventional rectangular architectures.1
The impact of this approach lies in dramatically accelerating design timelines.1 While conventional chip design requires six months to two years to complete, the AI-powered methodology can condense this cycle to approximately one week.1 This acceleration addresses an underlying urgency: computing demand is projected to grow by at least 10,000 times over the next five years.1 The founders advocate for a division of labor in which artificial intelligence handles execution-intensive tasks, freeing human engineers to concentrate on creative problem-solving and application development.1
评论
还没有评论，欢迎留下第一条。